A Memory Polynomial-Based Iterative Orthogonal Least Square Algorithm for Reconstructing Self-Interferences in Full-Duplex Communications

In in-band full-duplex (IBFD) communication systems, the digital self-interference cancellation (DSIC) is one of the most challenging problems. In this brief, we consider the DSIC from the perspective of compressed sensing and sparse reconstruction and propose an iterative orthogonal least square (IOLS) algorithm based on the memory polynomial-based self-interference canceller, which aims to select the most relevant polynomial terms (atoms) from the memory-polynomial dictionary and greedily acquires the sparse polynomial model coefficients. By incorporating the Gram-Schmidt orthogonalization and backward substitution into the IOLS, the sparse reconstruction of the digital self-interference (SI) is carried out iteratively avoiding direct calculation of the pseudo-inverse matrix in each selection step. Simulation results show that the proposed achieves up to 90% reduction in terms of the model complexity with regard to the conventional polynomial-based least square (LS) method while providing the comparable cancellation performance.